vortex_dict/builders/
mod.rs1use bytes::bytes_dict_builder;
5use primitive::primitive_dict_builder;
6use vortex_array::arrays::{PrimitiveVTable, VarBinVTable, VarBinViewVTable};
7use vortex_array::compress::downscale_integer_array;
8use vortex_array::{Array, ArrayRef};
9use vortex_dtype::match_each_native_ptype;
10use vortex_error::{VortexResult, vortex_bail};
11
12use crate::DictArray;
13
14mod bytes;
15mod primitive;
16
17#[derive(Clone)]
18pub struct DictConstraints {
19 pub max_bytes: usize,
20 pub max_len: usize,
21}
22
23pub const UNCONSTRAINED: DictConstraints = DictConstraints {
24 max_bytes: usize::MAX,
25 max_len: usize::MAX,
26};
27
28pub trait DictEncoder: Send {
29 fn encode(&mut self, array: &dyn Array) -> VortexResult<ArrayRef>;
31
32 fn values(&mut self) -> VortexResult<ArrayRef>;
33}
34
35pub fn dict_encoder(
36 array: &dyn Array,
37 constraints: &DictConstraints,
38) -> VortexResult<Box<dyn DictEncoder>> {
39 let dict_builder: Box<dyn DictEncoder> = if let Some(pa) = array.as_opt::<PrimitiveVTable>() {
40 match_each_native_ptype!(pa.ptype(), |P| {
41 primitive_dict_builder::<P>(pa.dtype().nullability(), constraints)
42 })
43 } else if let Some(vbv) = array.as_opt::<VarBinViewVTable>() {
44 bytes_dict_builder(vbv.dtype().clone(), constraints)
45 } else if let Some(vb) = array.as_opt::<VarBinVTable>() {
46 bytes_dict_builder(vb.dtype().clone(), constraints)
47 } else {
48 vortex_bail!("Can only encode primitive or varbin/view arrays")
49 };
50 Ok(dict_builder)
51}
52
53pub fn dict_encode_with_constraints(
54 array: &dyn Array,
55 constraints: &DictConstraints,
56) -> VortexResult<DictArray> {
57 let mut encoder = dict_encoder(array, constraints)?;
58 let codes = downscale_integer_array(encoder.encode(array)?)?;
59 unsafe { Ok(DictArray::new_unchecked(codes, encoder.values()?)) }
61}
62
63pub fn dict_encode(array: &dyn Array) -> VortexResult<DictArray> {
64 let dict_array = dict_encode_with_constraints(array, &UNCONSTRAINED)?;
65 if dict_array.len() != array.len() {
66 vortex_bail!(
67 "must have encoded all {} elements, but only encoded {}",
68 array.len(),
69 dict_array.len(),
70 );
71 }
72 Ok(dict_array)
73}